Truth Rules, Hoverflies, and the Kripke-Wittgenstein Paradox
Ruth Garrett Millikan
The Philosophical Review, Vol. 99, No. 3. (Jul., 1990), pp. 323-353.
Stable URL:
http://links.jstor.org/sici?sici=0031-8108%28199007%2999%3A3%3C323%3ATRHATK%3E2.0.CO%3B2-5
The Philosophical Review is currently published by Cornell University.
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained
prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in
the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/journals/sageschool.html.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org.
http://www.jstor.org
Mon Aug 27 11:51:31 2007
The Philosophical Review, Vol. XCIX, No. 3 (July 1990)
Truth Rules, Hoverflies, and the 

Kripke-Wittgenstein Paradox 

Ruth Garrett Millikan
[Tlhe sceptical argument that Kripke attributes to Wittgenstein, and
even the 'sceptical solution', are of considerable importance regardless
of whether they are clearly Wittgenstein's. The naturalistically inclined
philosopher, who rejects Brentano's irreducibility and yet
holds intentionality to be an objective feature of our thoughts, owes a
solution to the Kripke-Wittgenstein paradox.'
The challenge is a welcome one. Although I will argue that the
Kripke-Wittgenstein paradox is not a problem for naturalists
only, I will propose a naturalist solution to it. (Should the
Kripke-Wittgenstein paradox prove to be soluble from a naturalist
standpoint but intractable from other standpoints, that would, I
suppose, constitute an argument for naturalism.) Then I will show
that the paradox and its solution have an important consequence
for the theories of meaning and truth. The Kripke-Wittgenstein
arguments which pose the paradox also put in question Dummetťs
and Putnam's view of language understanding. From this view it
follows that truth rules must be "verificationist rules" that assign
assertability conditions to sentences, rather than "realist rules" that
assign correspondence truth conditions. The proposed solution to
the paradox suggests another view of language understanding, according
to which a speaker can express, through his language
practice, a grasp of correspondence truth rules. This will block one
route of Putnam's famous retreat from realism:
The point is that Dummett and I agree that you can't treat understanding
a sentence (in general) as knowing its truth conditions; because
it then becomes unintelligible what that knowledge in turn consists
in. We both agree that the theory of understanding has to be done
in a verificationist way ... conceding that some sort of verificationist
semantics must be given as our account of understanding. .. .I have
'Brian Loar, Critical Review of Saul Kripke's Wittgenstein on Rules and
Private Language, No& 19 (1985), p. 280.
323
RUTH GARRETT MZLLZKAN
given Dummett all he needs to demolish metaphysical realism . . . a
picture I was wedded to!2
(By "metaphysical realism" Putnam means, roughly, the traditional
correspondence theory of truth.) Elsewhere I have argued that the
distinction Putnam draws between "metaphysical realism" and "internal
realism" is illusory, that naturalist arguments for correspondence
truth are, inevitably, arguments for truth as correspondence
to theory-independent objects, and that there is nothing incoherent
in this notion of corre~pondence.~So in giving a naturalist
argument to show that grasping correspondence truth rules is no
more problematic than grasping verificationist ones, I take myself
to be defending the strongest possible kind of correspondence
theory of truth and the most flatfooted interpretation possible of
the truth-conditions approach to semantics.
The Kripke-Wittgenstein paradox, as Kripke explains it, is an
apparent dead end we encounter when trying to explain what it is
that constitutes a person's meaning something by a word. Kripke
takes addition as his central example: what constitutes my meaning
addition by "plus" or "+"? "Although I myself have computed
only finitely many sums in the past, the rule for addition determines
my answer for indefinitely many new sums that I have never
previously considered. This is the whole point of the notion that in
learning to add I grasp a rule: my past intentions regarding addition
determine a unique answer for indefinitely many cases in the
2Hilary Putnam, "Realism and Reason," in Meaning and the Moral
Sciences (London, England: Routledge and Kegan Paul, 1978). Michael
Dummetťs statement is in "What is a Theory of Meaning?" in Samuel
Guttenplan, ed., Mind and Language (Oxford, England: Clarendon Press
of Oxford University Press, 1975),pp. 97-139, and in "What is a Theory
of Meaning? (11)," in Gareth Evans and John McDowell, eds., Truth and
Meaning: Essays in Semantics (Oxford, England: Clarendon Press of Oxford
University Press, 1976), pp. 67- 137.
'In "Metaphysical Antirealism?" Mind 95 (1986), pp. 4 17-43 1;
reprinted in The Philosopher's Annual Vol. IX- 1986 (Atascadero, Calif.:
Ridgeview Publishing Co., 1988).
TRUTH RULES AND HOVERFLIES
f ~ t u r e . " ~What is it to "grasp" such a rule?What is it for me to have
grasped a rule that determines that 68 + 57 yields the answer 125,
in the case that I have never happened to add 68 to 57? No such
rule is determined merely by extrapolation from previous cases in
which I have applied "+" to pairs of numbers; there are always
infinitely many functions that accord with a given finite list of such
argument-argument-value trios. For example, the "quus" rule
might accord:
x quus y = x + y, if x,y < 57 

= 5 otherwise5 

Nor (and this is more obviously a Wittgensteinian theme) can we
suppose that my meaning addition by "+" consists in my having
given myself general directions for what to do when encountering
"+." To give myself general directions would be to lay down a rule
of procedure for myself. What then constitutes my meaning by this
set of instructions, by this laid-down rule, one procedure rather
than another? Certainly this set of instructions does not include a
thought of each of the infinitely many sums there are. And my
past performances when having this set of instructions in mind do
not exemplify a unique general procedure but many such possible
procedures. Supplementing the instructions with another set of
instructions explaining how to follow the first set leads only to a
regress. How then is the correct interpretation of the instructions
in my mind determined?
Changing the example, Kripke writes
It has been supposed that all I need to do to determine my use of the
word 'green' is to have an image, a sample, of green that I bring to
mind whenever I apply the word in the future. When I use this to
justify my application of 'green' to a new object, should not the sceptical
problem be obvious to any reader of Goodman? Perhaps by
'green', in the past I meant grue, and the color image, which indeed
was grue, was meant to direct me to apply the word 'green' to grue
objects always. If the blue object before me now is grue, then it falls in
the extension of 'green', as I meant it in the past. It is no help to
4S. Kripke, Wittgenstein on Rules and Private Language [hereafter
WORPL] (Cambridge, Mass.: Harvard University Press, 1982),p. 7.
51bid., p. 9.
RUTH GARRETT MILLIKAN
suppose that in the past I stipulated that 'green' was to apply to all
and only those things 'of the same color as' the sample. The sceptic
can reinterpret 'same color' as same schmolor, where things have the
same schmolor if. ..6
Now it is true that arguments of this sort take hold only if we
reject the possibility that intentionality is a sui generis feature given
to c~nsciousness.~We must assume that what comes before the
mind, whatever it is that enters or informs consciousness when one
means something, does not itself determine a use for itself, a purpose
for itself, a particular kind of connection that it is to have with
one's activities. Rather, whatever comes before the mind is, in this
respect, not different from any other item standing alone: "And
can't it be clearly seen here that it is absolutely inessential for the
picture to exist in his imagination rather than as a drawing or
model in front of him? . . Wittgenstein argues against the possibility
that intentionality is a sui generis feature, by showing, for each
of a series of cases, that the results of introspection when one
means, understands or is guided in accordance with rules, are not
the only or the final criteria that we use to determine what we
mean or when we understand or are being so guided. What lies
before consciousness does not determine its own significance;
knowing what one means is not a matter, merely, of apprehending
the contents of one's mind. In short, meaning is neither a state of
awareness nor an epistemological given. It does not occur encapsulated
within consciousness; it is not a state that simply shows its
content or its significance. If there is such a thing as meaning
something, say, meaning addition, its nature must lie in part in
what is not simply given to consciousness.
Nor is it merely because the object thought of or meant is external
to mind that meaning has an ingredient not given to consciousness.
Meaning to perform a mental activity like adding in the
61bid., p. 20. Kripke's ellipsis points at the end; Kripke's footnotes
omitted.
'Loar claims that Kripke has not demonstratedthat intentionality is not
this. Kripke's text does however contain several footnotes commentingon
the relevant arguments in Wittgenstein'stext. I mention these arguments
below.
Wittgenstein, Philosophical Investigations (New York, N.Y.: The Macmillan
Company, 1953),Paragraph 141.
- TRUTH
RULES AND HOVERFLIES
head, that is, having intentions about one's own thoughts, is fully
infected with this non-given ingredient. Thus the problem posed is
no different for the purest idealist than for the metaphysical
realist. Nor is it only "naturalistically inclined philosophers" who
need a solution to the Kripke-Wittgenstein paradox. It is anyone
who has been convinced by Wittgenstein to doubt Brentano-or,
say, convinced after Sellars to reject epistemological "givenness" in
all of its multifarious forms.
Could it be that the non-given ingredient that pins down what
rule I intend to follow for "+" is the disposition I have to proceed in
a certain way when encountering "+"? Setting aside the problem
of what Wittgenstein may have intended as an answer to this question,
surely Kripke is right to answer no. Kripke gives two main
reasons for his answer. First, people are in fact disposed to make
mistakes in arithmetic. Second, the addition function applies to
numbers of any magnitude, but "some pairs of numbers are simply
too large for my mind-or my brain-to grasp."g Nor will it help
to take into account dispositions I may have to correct myself or to
accept correction from others. Some of my dispositions are dispositions
to miscorrect myself. (I often do this when trying to add
long columns of figures.) And there are surely conditions under
which I would be disposed to accept miscorrection from others.
Kripke concludes, or he claims that Wittgenstein concludes, that
there is, indeed, no fact to the matter of what I mean by "+." This
conclusion is what I am calling the "Kripke-Wittgenstein paradox."1°
Wittgenstein, Kripke claims, offers only a "sceptical solugWORPL,pp.
26-27.
I0Kripkeplaces a great deal of emphasis on the failure to find anything
that "justifies" my proceeding as I do when I follow a rule, and he seems
to think of a "justification" as something that must be, by its very nature,
open to or within consciousness. Similarly: "Even now as I write, I feel
confident that there is something in my mind [italicsmine]-the meaning I
attach to the 'plus' sign- that instructs me [italicsKripke's] what I ought to
do in all future cases" (WORPL,pp. 21-22). And "The idea that we lack
'direcť access to the facts whether we mean plus or quus is bizarre in any
case. Do I not know, directly, and with a fair degree of certainty, that I
mean plus?" (WORPL, p. 40). Indeed, many passages in Kripke's essay
suggest that what bothers him the most is not that nothing seems to determine
what rule I am following,but that nothing before my mind determines
it. The feeling is conveyed that Kripke finds the real blow to be that the
intentionality involved in rule following does not reside within conscious-
RUTH GARRETTMZLLZKAN
tion" to this paradox, a solution that "begins . ..by conceding that
the sceptic's negative assertions are unanswerable."ll I propose to
offer a "straight solution" to this paradox, one that "shows that on
closer examination the scepticism proves to be unwarranted."12
Kripke distills the essence of the failure of dispositional accounts
to capture the nature of rule following thus:
A candidate for what constitutes the state of my meaning one function,
rather than another, by a given function sign, ought to be such
that, whatever in fact I (am disposed to) do, there is a unique thing
that I should do. Is not the dispositional view simply an equation of
performance and correctness?Assuming determinism, even if I mean
to denote no number theoretic function in particular by the sign '*',
then to the same extent as it is true for '+', it is true here that for any
two arguments m and n, there is a uniquely determined answer p that
I would give. (I choose one at random, as we would normally say, but
causally the answer is determined.) The difference between this case
and the case of the '+' function is that in the former case, but not in
the latter, my uniquely determined answer can properly be called
'righť or 'wrong'.13
The fundamental problem . . .is .. .:whether my actual dispositions
are 'righť or not, is there anything that mandates what they ought to
be?14
ness. If that is what Kripke takes to be the root "Wittgenstein paradox,"
then all will agree that Wittgenstein made no attempt to give a "straight
solution" to it. Nor will I. On the other hand, if that were the main paradox,
no one would ever have supposed that a dispositional account would
be a "straight solution" to it and Kripke's discussion of dispositional accounts
should have been placed not with his account of "Wittgenstein's
paradox" but with his discussion of "Wittgenstein's scepticalsolution," that
is, as an account of what Kripke believed this skeptical solution was not.
Margaret Gilbert suggests (in conversation) that one paradox may be
that meaning strikes one as being something that can be fully constituted
at a given time t, whether or not meaning is something that happens
within consciousness. And, to be sure, dispositions are usually taken to
exist at given times so that a dispositional account might be viewed as an
attempt at a straight solution to Gilberťs paradox. Gilberťs paradox, if
one finds it paradoxical, is another that I will not attempt to solve. I will
merely try to show how there is a fact to the matter of what I mean by " +."
l1W0RPL,p. 66. 

121bid. 

131bid.,p. 24; footnotes omitted. 

141bid.,p. 57. 

TRUTH RULES AND HOVERFLIES
The problem is to account for the normative element that is involved
when one means to follow a rule, to account for there being
a standard from which the facts, or one's dispositions, can diverge.
11. GENERALFORMOF THE SOLUTION
In the case of meaning, the normative element seems to be the
same as the purposive element: to mean to follow a certain rule is
to have as a purpose to follow it. Whether my actual dispositions
are "right" or "wrong" depends on whether they accord with what
I have purposed. The possible divergence of fact from a standard
is, in this case, simply the failure to achieve a purpose.
Now having as one's purpose to follow a rule might involve
having a representation of that purpose in mind, for example, in
one's language of thought. But as Wittgenstein observed, any such
representation would itself stand in need of interpretation. It
would stand in need of a prior rule governing how it was to be
taken, that is, how it was to guide one. And that one was to follow
this prior rule could not also be a represented purpose, not without
inviting a regress. T o understand what it is to have an explicit purpose
that one represents to oneself we must first understand what
it is to have a purpose the content of which is not represented.
Basic or root purposes must be unexpressed purposes.
"Intend" strongly suggests an explicitly represented purpose,
that is, a purpose that is thought about. So let me use the verb "to
purpose" (with a voiced "s"; yes, it is in the dictionary) to include
this more basic way of having a purpose. We can then put matters
this way: root purposing is unexpressed purposing; our job is to
discover in what this purposing consists. Let us also distinguish
among three ways of conforming to a rule: (1) merely coinciding
with a rule (this is the way in which we conform to "quus" rules
and to rules to which we have mere dispositions to conform), (2)
purposefully following an explicit or expressed rule, and (3) purposefully
conforming to an implicit or unexpressed rule. Way (3)
involves having an unexpressed purpose to follow a rule and succeeding
in this purpose. It is the same as displaying a competence in
conforming to the unexpressed rule or displaying an ability to conform
to it. Another way to explain our task, then, is to say that we
need to learn what a competence in conforming to an unexpressed
rule consists in, and how it differs from a mere disposition to coincide
with the rule.
RUTH GARRETT MZLLZKAN
My thesis will be that the unexpressed purposes that lie behind
acts of explicit purposing are biological purposes; a competence to
conform to an unexpressed rule is a biological competence. By a
biological purpose I mean the sort of purpose the heart has, or
those of the eyeblink reflex, and the human brain. The purposes
of these are functions that they have historically performed which
have accounted for their continued proliferation. Biological purposes
are, roughly, functions fulfilled in accordance with evolutionary
design. It does not follow that capacities to perform biological
functions are, in general, innate. For example, it is surely in
accordance with evolutionary design that the newly hatched chick
follows its mother about, but the chick is not born with that disposition.
It is not born knowing which is its mother, but must imprint
on her first. Yet the imprinting, and hence the following,both take
place in accordance with evolutionary design. (Later in this paper I
will devote considerable space to clarifying how even quite novel
biological purposes can emerge as a result of experience and
learning.)l5
Suppose that explicit intending involves something like representing,
imaging, or saying something to oneself and then using,
or reacting to, or being guided by this representation in a purposeful
way, that is, in a way that expresses a competence. My
thesis, then, is that the purpose that informs this reacting, that
makes it into a competence, is a biological purpose. Similarly, if
knowing a language involves having a competence in following
certain rules for construction and interpretation of sentences, the
purpose that informs this competence, I will argue, is a biological
purpose.
III. PURPOSIVERULEFOLLOWING; TO FOLLOWCOMPETENCE
A RULE
Let me begin with a very simple example of an organism that
displays a competence in conforming to a rule. According to the
biologists Collett and Land,
15Fulldetails of the notion of biological function that I rely on in this
paper are given under the label "proper functions" in my Language,
Thought, and Other Biological Categories [hereafter LTOBC] (Cambridge,
Mass.:Bradford BooksIThe MIT Press, 1984), Chapters 1-2. See also my
"In Defense of Proper Functions,"Philosophy of .hence 56 (1989), pp.
288-302.
TRUTH RULES AND HOVERFLIES
Males of many speciesof hoverfly spend much of the day hovering
in one spot, thus keeping their flight muscles warm and primed so
that they are ready to dart instantly after any passing female that they
sight. This chasing behavior is on such a hair-trigger that all manner
of inappropriate targets elicit pursuit (pebbles, distant birds, and
midges so small as to be scarcely visible to a human observer) as well
as a very occasionalfemale.Although selective pressures have favored
a speedy response above careful evaluation of the suitability of the
target ...the response itself is precisely tailored to optimize the capture
of objects which are roughly the same size and speed as a con-
specific.16
Rather than turning toward the target in order to track it, the
hoverfly turns away from the target and accelerates in a straight
line so as to intercept it. Given that (1) female hoverflies are of
uniform size, hence are first detected at a roughly uniform distance
(about .7 m), (2) females cruise at a standard velocity (about
8m/sec), and (3) males accelerate at a constant rate (about
30-35m/sec2), the geometry of motion dictates that to intercept
the female the male must make a turn that is 180" away from the
target minus about 1/10 of the vector angular velocity (measured
in degrees per second) of the targeťs image across his retina. The
turn that his body must make, given as a function of the angle off
center of the targeťs image on his retina, equals the (signed) angle
of the image minus 1/10 its vector angular velocity, plus or minus
180".According to Collett and Land, whether it is dried peas, male
hoverflies, female hoverflies or flying blocks of wood that he
spots, that is exactly the rule to which the hoverfly conforms.
Taking note that this rule is not about how the hoverfly should
behave in relation to distal objects, but rather about how he should
react to a proximal stimulus, to a moving spot on his retina, let us
call this rule "the proximal hoverfly rule."
I have chosen the proximal hoverfly rule as my first example of
rule following because it seems so unlikely that the hoverfly calculates
over any inner representation of this rule in order to follow it.
Rather, the hoverfly has an unexpressed biological purpose to
conform to this rule. That is, the hoverfly has within him a genetically
determined mechanism of a kind that historically proliferated
16"HowHoverflies Compute Interception Courses,"Journal of Comparatzue
Physiology 125 (1978), pp. 191-204.
RUTH GARRETT MILLZKAN
in part because it was responsible for producing conformity to
the proximal hoverfly rule, hence for getting male and female
hoverflies together. This mechanism may account for various
other dispositions of the hoverfly, for example, causing him to attract
predators by his conspicuous darting movements, or causing
characteristic uniform mathematically describable patterns to play
on his retina as he turns after the female. But mentioning theselatter
dispositions does not help to explain why the mechanism has
survived, why it has proliferated in the species. Conformity to
the proximal hoverfly rule, on the other hand, has helped to explain
the reproductive success of (virtually) every ancestor
hoverfly, hence to explain the continued presence of the mechanism
in the species. Conformity to the proximal hoverfly rule,
then-not attracting predators or producing certain patterns on
the retina-is a biological purpose of this mechanism, hence of the
hoverfly. For similar reasons, a biological function of the heart is
to pump blood but not also, say, to make a jazzy sound, and a
biological function of the eyeblink reflex is to cover the eyes momentarily,
but not also to swing the eyelashes in a graceful arc
away from entanglement with the eyebrows, nor to point with
them at the navel.
The hoverfly displays a competence in conforming to the proximal
hoverfly rule when his coinciding with it has a "normal explanation,"
that is, an explanation that accords with the historical
norm. That his behavior coincides with the rule must be explained
in the same way, or must fit the same explanation schema, that
accounted in the bulk of cases for the historic successes of his ancestors
in conforming to the rule. Presumably this normal explanation
makes reference to the way the hoverfly's nervous system is
put together, how it works, how it is hooked to his retina and
muscles, etc. If the hoverfly ends up coinciding with the rule not
because his nerves and muscles work in a normal way but only
because the wind serendipitously blows him around to face the
right direction, he fails to express a competence.17
Notjust anything a human effects is a human action. Effects that
are actions must be intended, or at least foreseen, and must be
"For a full discussion of normal ("Normal") explanations for performance
of proper functions, see LTOBC, Chapters 1 and 2, and my "Biosemantics,"
TheJournal of Philosophy 86 (1989),pp. 281-297.
TRUTH RULES AND HOVERFLIES
generated from intentions in a normal way. Effects of human
bodily movements that are not actions are called "accidental."
Similarly, notjust any process that originates in an animaľs organs
or behavioral systems is a biological activity. Biological activities are
only those that express competencies. They correspond to normally
fulfilled biological purposes, that is, to what the animal does
in accordance with evolutionary design. Conversely, behavior that
fails to express a competence corresponds to what an animal effects,
biologically, only by accident. Thus the hearťs saying pit-apat,
the eyelashes' moving away from the eyebrows in a graceful
arc and the hoverfly's coinciding, but due only to the wind, with
the proximal hoverfly rule are not biological activities, but biological
accidents.
To say that a given male hoverfly has a biological purpose to
conform to the proximal hoverfly rule is very different from
saying either that-he himself has a history of having conformed to
it (perhaps he has just reached adolescence) or that he has a disposition
to conform to it. The normal hoverfly has a disposition to
dart off when it sees a flying bird-and also a disposition to
squash when stepped on-but these dispositions do not correspond
to biological purposes or to competences. Conversely, male
hoverflies that are crippled or blind have no disposition to conform
to the proximal hoverfly rule, but still it is one of their biological
purioses to do so. AS male members of the hoverfly
species, conforming is the biological norm, the standard for
them.ls
To say that the hoverfly has as a biological purpose to follow the
proximal hoverfly rule is also quite different from saying that this
rule is the only rule that fits all past instances of hoverfly turns,
say, that resulted in hoverfly procreation. Suppose it were so that
never in history had a male hoverfly spotted a female that happened
to approach him at such an angle as to produce an image on
his retina with a clockwise angular velocity between 500" and 510"
per second. Then the proximal quoverfly rule, "If the vector angular
velocity of the targeťs image is not counterclockwise and between
500" and 510" per second, make a turn that equals the
(signed) angle of the image minus 1/10 its vector angular velocity,
180n the proper functions of imperfect members of a biological category,
see LTOBC, Chapters 1 and 2.
333
RUTH GARRETT MZLLIKAN
plus or minus 180"; at ease otherwise," fits all past actual cases of
successful female encounters. But it is not a rule the hoverfly has
as a biological purpose to follow. For it is not because their behavior
coincided with that rule that the hoverfly's ancestors managed
to catch females, hence to proliferate. In saying that, I don't
have any particular theory of the nature of explanation up my
sleeve. But surely, on any reasonable account, a complexity that
can simply be dropped from the explanans without affecting the
tightness of the relation of explanans to explanandum is not a
functioning part of the explanation. For example, my coat does not
keep me warm because it is fur-lined and red, nor because it is furlined
in the winter, but just because it is fur-lined. (True, I am
making the assumption that the qualifications and additions that
convert the proximal hoverfly rule into the proximal quoverfly
rule are objectively qualifications and additions rather than simplifications.
This assumption rests upon a metaphysical distinction
between natural properties and kinds and artificially synthesized
grue-like properties and kinds or, what is perhaps the same, depends
upon there being a difference between natural law and
mere de facto regularity. But my project is to solve the KripkeWittgenstein
paradox, not to defend common-sense ontology. Nor
should either of these projects be confused with solving
Goodman's paradox.)lg
To say that the hoverfly has as a biological purpose to follow the
proximal hoverfly rule is also quite different fro111saying that this
rule is the only rule that fits the actual dispositions of nornlal
hoverflies or of past hoverflies that managed to procreate. Suppose
that, given the principles in accordance with which the
hoverfly's tur11-arlgle-deternli~lirlgdevices work, engineering constraints
necessitated a nlecharlisnl normal for hoverflies with a
blind spot for clockwise angular velocities between 500" and 510"
per second. These particular velocities produce no reaction at all
on the part of the male. Then the same proxinlal quoverfly rule
'g(;oodman's paradox is a paradox in epistemology. Kripke, on the
other hand, is concerned not about how we could know or discover what
someone means by "plus" but about what this determinate meaning co.r~~~,sts
in. Note too that assuming common-sense ontology does nothing, by itself,
toward solving Goodman's paradox, which concerns how we can know or
reasonably guess which entities are the basic ontological ones, .su~@posir~g
there to be such.
TRUTH RULES AND HOVERFLIES
mentioned above fits the actual dispositions of all normal hoverflies,
but it still would not be a rule that the hoverfly has as a biological
purpose to follow. The hoverfly's biological purposes include
the expression only of dispositions that have helped to account
for the proliferation of his ancestors. By hypothesis, the
disposition to rest at ease when the targeťs image is counterclockwise
and between 500" and 510" per second did not help the
hoverfly's ancestors to propagate. It was only the times that the
proximal hoverfly rule was obeyed that the ancestors procreated.
So the hoverfly resting at ease behind his blind spot is not displaying
a competence. It is conformity to the proximal hoverfly
rule, not the quoverfly rule, that he biologically purposes, even if
normal hoverflies are not especially accurate in fulfilling this natural
purpose, in conforming to this ideal.
IV. PROXIMALVS. DISTALRULES
My plan, as I have indicated, is slowly to make plausible the
claim that the normative element that is involved when one means
to follow a rule is biological purposiveness. Meaning to follow a
rule differs from having a disposition to coincide with a rule, in the
same way that the hoverfly's biologically purposing to follow the
proximal hoverfly rule differs from having a disposition to coincide
with it. That is how I aim to solve the Kripke-Wittgenstein
paradox concerning what constitutes rule following. At the same
time, however, I wish to build a case that language understanding
or language competency is competency in the biological sense.
And I wish to argue that it is possible to have a biological competence
to follow correspondence truth rules, hence that a "realist"
theory of language understanding is possible on the biological
model. To gain this latter end, we need to discuss distal as well as
proximal rules.
Conforming to the proximal hoverfly rule is a means, for the
hoverfly, of following a less proximal, or more distal rule: "If you
see a female, catch it." Call this "the distal hoverfly rule." To say
that conformity to the proximal hoverfly rule is a means to conformity
to the distal rule is the same as to say that the mechanism
that has historically accounted for the overwhelming majority
of ancestor hoverflies' successes at conformity to the distal
hoverfly rule begins with conformity to the proximal rule. That is,
the normal explanation for conformity to the distal rule contains
RUTH GARRETT MZLLIKAN
the specification that the hoverfly first conform to the proximal
rule.
Now whether the hoverfly succeeds in following the proximal
hoverfly rule depends, for the most part, only upon whether his
insides are working right, that is, on whether he is a normal
healthy member of his species. But whether, or how often, he
manages to conform to the distal hoverfly rule depends upon
more. It depends upon conditions that are outside his body and
over which he has no control, such as how hard the wind is
blowing, whether the females that pass by are in fact of normal
size, traveling at the normal speed and, perhaps, whether they are
willing. Without doubt, then, hoverflies are worse at conforming
to the distal than to the proximal hoverfly rule. That is, their competence
or ability to conform to the distal rule is less reliable than
their competence or ability to conform to the proximal rule. But
that the hoverfly may not be very reliable in his conformity to the
distal hoverfly rule bears not at all upon whether it is one of his
biological purposes to conform. Compare: it is a biological purpose
of the sperm to swim until it reaches an ovum. That is what it has a
tail for. But very few sperm actually achieve this biological end
because ova are in such short supply. Reaching an ovum is a purpose
of the sperm since it is only because ancestor sperm reached
ova that they reproduced, thus proliferating the tail. Similarly, it
was only when ancestor hoverflies conformed to the distal
hoverfly rule that they became ancestors.
Turning the coin over, the hoverfly is very reliable in his coincidence
with this "overkill rule": "Dart off after everything that flies
by you subtending about .5" on your retina, whether iťs male, female,
animate or inanimate, bird, plane or Superman." But this
overkill rule does not correspond to any biological purpose of the
hoverfly. True, conforming to the proximal hoverfly rule is one of
the hoverfly's biological purposes, and conforming to this rule will
result in his coinciding with the overkill rule if there are objects
other than female hoverflies flying about him (even if there are
not). But it is not coinciding with the overkill rule that has helped
to account for hoverfly proliferation. Only the times when the
distal hoverfly rule was obeyed did hoverfly ancestors procreate.*O
'OMore precisely, only the distal hoverfly rule would be mentioned in
giving a "most proximate normal explanation" of the function of the
TRUTH RULES AND HOVERFLIES
It is conformity to the distal hoverfly rule that explains the ancestor
hoverflies' successes. As the hoverfly chases after a distant
bird, he expresses no competence except, of course, competence to
conform to the proximal hoverfly rule. Conformity to the distal
hoverfly rule, not to the overkill rule, is what he biologically
purposes, though at the moment he is accidentally, that is, nonbiologically,
doing something else.
That is how purposes inform the rule-following behavior of the
hoverfly, how norms, standards, or ideals apply to his behaviors,
hence how the hoverfly comes to display competences or abilities
to conform to rules rather than mere dispositions to coincide with
them.21But the unexpressed rules that humans purposively conform
to, at least most of those that they purposively conform to
when using inner or outer language, are not rules that they are
genetically hard-wired to follow, but rules that they have learned.
How then can humans biologically purpose to follow such rules?
Before turning directly to the problem of human rule following,
let us examine a more simple case of learned biological purposes,
of learned competence-the case of a simpler animal that learns to
follow rules.
V. LEARNEDOR DERIVEDRULESAND COMPETENCES
If a rat becomes ill within a few hours after eating a specific
food, it will later shun all foods that taste the same. For example, if
the rat eats soap and soon becomes ill, thereafter it will refuse to
eat soap. Although the rat may have dragged certain nesting materials
home or explored new territory just before becoming ill, it
will not on that account shun that kind of nesting material or that
territory. Nor will it shun foods that merely look the same or that
are found in the same place as the food eaten prior to illness.22It
hoverfly's turning mechanism. See LTOBC, and the discussion o f "normal
explanations" in m y "Biosemantics."
21Notice that it is the reference to evolutionary history that has been
doing all o f the work in explaining how norms come to apply to the activities
o f an animal, in explaining how there can be a standard from which
the facts o f individual behavior diverge. I defend the position that function
always derives from history in "In Defense o f Proper Functions."
22The reference is to studies by John Garcia. A bibliography o f his
papers may be found in The American Psychologist 35 (1980), pp. 41-43.
RUTH GARRETT MZLLZKAN
thus appears that a quite specific mechanism is harbored in the rat,
a proper function of which is to produce conformity to the specialized
rule "If ingestion of a substance is followed by illness, do not
ingest any substance with that taste again." Call this rule the
"proximal rat rule." Clearly, following the proximal rat rule is a
biological means to following a more distal rat rule, say, "Do not
eat poisonous substances"; helping to produce conformity to this
rule is a further proper function of the relevant inborn mechanisms
in the rat.
Now the proximal rat rule, like the proximal hoverfly rule, tells
the animal what to do given certain experiential contingencies.
There is a difference, however, in the normal manner of executing
these two rules. When the hoverfly conforms to his rule, nothing
in his body undergoes a permanent change, but this is not so in the
case of the rat. Suppose, for example, that the rat hasjust become
ill after eating soap. In order to conform to the proximal rat rule,
in order to avoid henceforth what tastes like soap, the raťs nervous
system must first conform to certain preliminary "rules," rules that
dictate that a certain sort of permanent change take place in it.
The rat, we say, must "learn" in order to conform to his rule. But
the fact that the raťs evolutionary history dictates that it is normal
for him to undergo learning in order to follow his rule rather than
following it directly does not affect the biological status of the rule.
That he should follow his rule is one of his biological purposes for
exactly the same reason that the hoverfly's rule following is biologically
purposed. Conformity to the rat rule is what his ancestor
rats had in common in those cases in which possession of the relevant
inborn mechanisms aided them to flourish and proliferate, so
it is what the mechanism, hence the rat, biologically purposes.
Now the rat that conforms to the proximal rat rule, if he ever
becomes ill after eating, ends by conforming to a derived proximal
rat rule, say, the rule "Do not eat what tastes like soap." Indeed, if
a rat becomes ill after eating soap, it immediately becomes one of
his biological purposes to follow the rule "Do not eat what tastes
like soap." For that he is to follow this derived rule is logically entailed
by the proximal rat rule plus the premise that he has in fact
become ill after eating soap. Similarly, the hoverfly that currently
has an image of appropriate size traversing his retina at a 60"angle
with an angular velocity of 100"per second currently has as a biological
purpose to make a turn of 130".Notice that the hoverfly
- TRUTH
RULES AND HOVERFLIES
has this biological purpose quite independently of whether or not
any hoverfly has ever been in exactly this experiential position before.
It is theoretically possible, even if unlikely, that no hoverfly
has ever had exactly this biological purpose before. This is similar
to our rat who is sick after eating soap. It is now one of his biological
purposes to follow the derived proximal rat rule "Do not
eat what tastes like soap" even if it should be true that no other rat
in history has ever become sick after eating soap, hence true that
no rat in history has ever had this particular biological purpose
before.
In this manner, animals that learn can acquire biological purposes
that are peculiar to them as individuals, tailored to their own
peculiar circumstances or peculiar histories.23Although biological
purposes are functions fulfilled in accordance with evolutionary
design, they need not be innately given purposes. Similarly, biological
competences need not be innate. A proper biological activity
of an animal can be something that experience has prompted
or "taught" the animal to do, experience coupled with an innate
mechanism for being guided to learn by experience.
Nor is there need for such mechanisms to be as specialized as the
mechanism that conforms the rat to the proximal rat rule. Not
long ago many learning theorists believed that all animal learning
took place in accordance with principles that were not species
specific but universal. Suppose that this were true. Suppose that
every species learned in accordance with the principles of one
person's favorite general theory of operant conditioning, so that
no reference to the particular evolutionary niche of a species was
ever needed to explain how its learning mechanisms had historically
enhanced its fitness. Then there would have to be some rarefied
hypergeneral explanation of how and why these learning
principles worked. Such an explanation might make reference, for
example, to specific principles of generalization and discrimination
used in differentiating stimuli and in projecting what is to
count as "the same" behavior again, that is, reference to universal
proximal rules followed during learning. It would have to tell how
23Amuch more detailed discussion of "derived proper functions" may
be found in LTOBC, Chapter 2, and in my "Thoughts Without Laws:
Cognitive Science With Content," The Philosophical Review 95 (1986), pp.
47-80.
RUTH GARRETT MZLLZKAN
and why these particular ways of generalizing and discriminating
effected, often enough, isolation or zeroing in on sufficiently reliable
causes of reinforcement, and in what universal manner (!)
reinforcers are connected with the well-being of animals. Thus it
would tell how possession of the universal mechanism had normally,
that is, historically, enhanced fitness in animals generally.
Specific applications of this general explanation schema to individual
animals in individual circumstances would then determine
which among the various effects of their motions were the proximal
and distal biological activities of these individuals, as they
learned and applied their learning. Such applications would determine,
for example, what specific reliable causes of reinforcement
were purposefully being zeroed in on by particular animals at particular
times, that is, what these animals were "trying" to learn and,
after they learned it, what the specific goals of their learned behaviors
were.
Now it is imporant to note that to fulfill a biological purpose is
not always to take a step towards flourishing or propagating; it is
not always good for an animal to fulfill its biologically determined
goals. For example, a rat might come to have as a biological purpose
to follow the derived rule "Do not eat what tastes like soap"
even if it were true (I suspect it is true) that soap does not make rats
sick or does not poison them. Suppose, rather, that the rat eats
soap and then becomes ill due to a bout with Rattus enteritis. Still,
in order to conform to the proximal rat rule, he must now conform
to the derived proximal rule "Do not eat what tastes like
soap," for this derived rule is entailed by the proximal rat rule
given his situational experience. Yet following this derived rule
may, in fact, have no tendency to bring him into conformity with
the more distal rat rule "Do not eat poisonous substances." So it
can happen that the rat acquires a biological purpose and acquires
a competence to conform to a derived rule which does not further
the end that is this rule's own raison ďttre. Indeed, the rat could
acquire a derived purpose and a competence to behave in a
manner that was actually detrimental to him, say, a competence to
follow the rule "Do not eat what tastes like corn" when, in fact,
unless he eats corn, given his circumstances, he will starve. Compare:
the hoverfly, dutifully conforming to the proximal hoverfly
rule (the rule that tells how he is to react to a moving image on his
retina) may thereby dart off after a bird, who would not otherwise
TRUTH RULES AND HOVERFLIES
have spotted the hoverfly, hence would not have eaten him. Thus it
is that an individual may have a biological purpose and a competence
to follow a derived rule that has no tendency to further
the interests either of the individual or of his species and, more
specifically, no tendency to produce conformity to more distal
rules toward which following it was, biologically, supposed to be a
mea11s.2~
What an animal is doing in accordance with evolutionary design
need not be anything that any member of its species has ever done
before. And it need not be anything that is good for the animal to
do. So surely it need not be anything that common sense would call
"natural" for it to do. Consider a circus poodle riding a bicycle. It
is performing what common sense would call a most "unnatural"
act. Yet it is one of the dog's biological purposes to perform that
act. Biologically, the (typical circus) dog's distal action is procurement
of his dinner. The dog harbors within him an intricate mechanism,
operating in accordance with certain largely unknown but
surely quite definite and detailed principles, in accordance with
which dogs have been designed to develop perceptual, cognitive
and motor skills and to integrate them so as to effect procurement
of dinner in their individual environments. Living in an unusual
environment, the circus dog acquires unusual purposes and competences
when he applies his "dog rules" to his environment. But,
although he may be making the audience laugh by accident, he is
certainly not balancing on that bicycle by accident. He is balancing
purposefully or in accordance with evolutionary design-in
accordance with another application of the same general principles
that procured his ancesters' dinners during evolutionary his-
tory.
VI. HUMANRULEFOLLOWING
Humans are very sophisticated creatures, so we tell ourselves.
We not only learn but learn new ways to learn, develop new concepts,
and so forth. Further, much of our behavior results notJust.
from learning but from theoretical and practical inference. But.
there must still be a finite number of inborn mechanisms, oper2
4 F ~ rfurther details on conflicting proper functions, see Ly'OB(;,
Chapter 2, and rrly "Thoughts Without Laws."
311
RUTH GARRETT MILLIKAN
ating in accordance with a finite number of natural principles,
having a finite number of biologically proper functions, that account
for our dispositions to do these things. Coordinately, there
must be a finite number of proximal and distal "Homo sapiens
rules" that we have as biological purposes to follow, and there
must be mechanisms to implement these rules built into the basic
body and brain of normal persons.
Consider then any bit of human behavior produced by biologically
well-functioning behavior-regulating systems, by systems that
are not broken or jammed. (Behavior that results from malfunction
is, of course, overwhelmingly unlikely to bear fruit of any interesting
kind.) There will be a way of describing this behavior that
captures its aspect as a biolopcal activity, a description that tells
what proximal and distal biological purposes, and what biological
competences if any, the behavior expresses. This will be so even if
the behavior is totally unique, or systematically self-destructive, or
not "natural" by any common-sense standards. But of course there
will also be numerous ways of describing the behavior that fail to
express its biological purposiveness, many "quus-like" descriptions
-as "pointing toward the navel with the eyelashes" quus-describes
the eyeblink reflex. So the question arises, what is the relation of
ordinary human purposes, of human intentions and meanings, to
biological purposes? Are descriptions of human intentional actions
quus-descriptions from the standpoint of evolutionary design? Do
ordinary human intentions merely, accidentally, cohabit with biological
purposes?
Surely a naturalist must answer no. Ordinary human purposes,
ordinary intentions, can only be a species of biological purpose. To
suppose otherwise would be to suppose that the whole mechanism
of human belief, desire, inference, concept formation, etc., the
function of which culminates in the formation and execution of
human intentions, is, as functioning in this capacity, an epiphenomenon
of biology, an accidental by-product of systems that nature
designed for other purposes. And what would these other purposes
be?25
This accords with conclusions we reached earlier on the nature
of explicit intentions. Explicitly meaning or intending, if this re25Fora
more detailed defense of this claim, see my "Biosemantics."
TRUTH RULES AND HOVERFLIES
quires representing what one intends, presupposes a prior purposing:
purposing to let the representation guide one in a certain
way. This is true whether we are talking about representation in an
inner medium, say, in a "language of thought," or representation
in a public medium-talking, say, about the use of "plus." But this
prior purposing cannot be analyzed as the original explicit purposing
was analyzed without regress. Rather, a prior unexpressed
purposing must be assumed. The reasonable conclusion seems to
be that ordinary explicit intending rests on biological purposingbiologically
purposing to be guided by, to react this way rather
than that to, one's representations. Whether this biological purposing
is innate (compare Fodor's version of the "language of
thought") or whether it is derived via learning, mechanisms of
concept formation, etc., it must ultimately derive its content from
the details of our evolutionary history.
So unless doing arithmetic results from a total breakdown of the
cognitive systems (in which case there may be nothing you purpose
when you encounter "plus": how you react to it is accidental under
every description) then whatever you mean to do when you encounter
"plus," that content has been determined by your experience
coupled with evolutionary design. But, reasonably, whatever
you mean by "plus" is the same as what other people mean who are
endowed with the same general sort of cognitive equipment and
have been exposed to the same sort of training in arithmetic. This
meaning has been determined by the application of Homo sapiens
rules of some kind to experience. It is likely that these are extremely
abstract general purpose Homo sapiens rules, in accordance
with which human concept formation takes place, and it is likely
that the explanation of the efficacy of these rules makes reference
to very deep and general principles of ontology. But it is not my
task to speculate about the precise form these Homo sapiens rules
take, or about how the experience of standard training in arithmetic
elicits from them the capacity to mean plus. Speculation
about the specific forms that our most fundamental cognitive
capacities take is the psychologisťs job.26
26B~t,people still persist in asking, How do you know that we really do
end up meaning plus by "plus"? How do you know we don't mean quus?
Because if we meant quus then "plus" would mean quus, and the way to
say that we all meant quus would be "we all mean plusn-which is what I
343
RUTH GARRETT MZLLZKAN
I believe that these considerations constitute, albeit in very 

rough and broad outline, the solution to the Kripke-Wittgenstein 

paradox. 

VII. TRUTHRULES:VERIFICATIONISTOR CORRESPONDENCE?
I have sketched a theory about meaning in the sense of purposing-both
expressed and unexpressed purposing. It remains
to connect this theory with the theory of semantic meaning.
Truth rules are rules that project, from the parts and structure
of sentences in a language, the conditions under which these sentences
would be true. Such rules express, of course, an aspect of
the meaning of the sentences. The question is whether the conditions
referred to by truth rules are to be understood in a "realist"
way as correspondence truth conditions, or in a "verificationist"
way as assertability conditions. Dummetťs concern about truth
rules is this: whatever connection there is between sentences and
that which determines their truth has to be a connection that is
established via the actual employment of the language. Whatever
form truth rules take, realist or verificationist, the practical abilities
of speakers who understand a language must reflect these rules,
indeed, must determine their content. Hence an analysis of the
structure of the abilities required for language use and understanding
should reveal the kind of rules truth rules are. But,
Dummett argues, the only truth rules we could possibly exhibit a
practical grasp of are verificationist truth rules.
In Section VIII below, I will claim that Dummetťs argument
hangs on treating language abilities or competencies, hence the
following of language rules, as mere dispositions, or alternatively
(perhaps), as taking place wholly within consciousness, and I will
add to the arguments already piled up by Wittgenstein and Kripke
against the adequacy of this sort of treatment. In the present section,
however, I wish to propose a positive thesis. My claim will be
that if we interpret rule following and, in general, purposes and
competencies in the biological way, then we can see how, on the
said. Compare Donald Davidson, "Knowing One's Own Mind," Proceedings
and Addresses of the American Philosophical Association 60 (1987), pp.
441-458, and Tyler Burge, "Individualism and Self Knowledge," Journal
of Philosophy 60 (1988), pp. 649-663.
344
TRUTH RULES AND HOVERFLIES
contrary, reference to correspondence truth rules might easily fall
out of an analysis of language competence.
We begin by observing that whatever the content of truth rules
may be, realist or verificationist, the intent or purpose of anyone
engaged in making sincere assertions in a language must be to
conform their sentences to these rules. The sincere speaker purposes
to make assertions that are true. It follows that the way that
the actual practice of a language embodies truth rules is that these
are the rules in accordance with which the competent speaker (or
thinker), when sincere, purposes to make (or think) assertions.
These are rules that he is, as it were, trying to follow insofar as he is
sincerely speaking (or thinking) that language. On the bottom layer
at least (perhaps the layer that governs the language of thought)
these rules must of course be unexpressed rules. But precisely because
truth rules are at bottom unexpressed rules, introspection
can give us no handle on what kind of rules they are. Rather, it is
necessary to develop a theory about truth rules, an explanatory hypothesis
about what rules we are purposing to follow when we
make sincere assertions.
Assuming a biological standpoint, the question whether truth
rules are realist or verificationist can be expressed by asking how
"proximal" vs. "distal" truth rules are. The proximal hoverfly rule
was a rule about how the hoverfly was to respond to a moving
image on his retina, that is, roughly, to sensory stimulations. The
distal hoverfly rule was a rule about how the hoverfly was to end
up interacting with his more removed environment, namely, with
females that entered his life at a distance. "Verificationist" truth
rules, as Putnam and Dummett envision these, would be rules that
governed responses to prior thoughts and, as Dummett has put it,
"bare sense experiences," hence would be proximal rules.*'
"Realist" or correspondence rules, on the other hand, would for
the most part be distal rules, rules that governed the manner in
which assertions were to correspond to affairs that lie, very often,
well beyond the interface of body and world. Convinced by
Wittgenstein and Kripke that purposing to follow a rule is not
something encapsulated within consciousness, we are not compelled
to suppose that truth rules have to be rules about what is to
happen either in the mind or at the interface between mind, or
27"Whatis a Theory of Meaning? (11)," p. 111.
345
RUTH GARRETT MILLIKAN
body, and world. So let us ask what it would be like if truth rules
were distal correspondence rules.
The first thing to note is that if truth rules were distal rules they
would surely have to be backed by proximal rules, rules about how
to respond to our thoughts (inference) and to the immediate fruits
of our perceptual explorations (perceptual judgment). They
would have to be backed by rules that determined assertability conditions,
the innermost of these conditions being within the mind or
brain or at the interface of mind or brain and world. Call these
back-up rules "proximal assertability rules." Proximal assertability
rules would concern the most proximal conditions under which we
should say or think certain things. Conformity to these rules would
have, as a biological purpose, to effect conformity to distal rules,
that is, to correspondence truth rules. These truth rules would
concern distal conditions under which we should say or think certain
things. The truth rules might imply directives with this sort of
form: if you have reason to speak (think) about the weather in
Atlanta, say (think) "It is snowing in Atlanta" when and only when
it is snowing in Atlanta; if you have reason to speak (think) about
the color of snow, say (think) "Snow is white" if and only if snow
is white. For a simple biological model here, compare worker
honeybees. They (biologically)purpose to follow rules of this kind:
when dancing, angle the axis of your dance 10"off the vertical if
and only if there is a good supply of nectar 10" off a direct line
from hive to sun. (Proposals concerning how humans might learn
how to (purpose to) conform to distal correspondence truth rules
are detailed in LTOBC.)28
Conforming to the proximal hoverfly rule and the proximal rat
rule often fails to bring hoverflies and rats into conformity to the
distal hoverfly and rat rules. Similarly, conforming to proximal
assertability rules might often fail to bring humans into conformity
to truth rules. One can unknowingly say what is false even though
one has good evidence for what one says. And one frequently fails
to say what is true, indeed, to say anything at all, because one lacks
any evidence at all, either for or against. Also, whether conformity
to the proximal hoverfly and rat rules helps to produce conformity
to the distal hoverfly and rat rules on this or that occasion
28Chapters9, 17 and 18.
TRUTH RULES AND HOVERFLIES
often depends upon factors in the hoverfly's or raťs external environment
over which it has no control. Similarly, whether conformity
to proximal assertability rules would bring us into conformity
to truth rules in this case or that might depend upon
factors over which we had no control. For example, circumstances
responsible for most perceptual illusions are circumstances outside
the observer which, normally, he neither controls nor needs to
control. Nor is not having enough evidence either to affirm or to
deny a proposition typically something that it is within one's control
to remedy. The principles in accordance with which biological
devices perform functions that are proper to them always refer, in
the end, to conditions external to these devices. These are conditions
that have historically been present often enough to enable a
critical proportion of ancestors of those devices to perform these
functions, or to perform them a critical proportion of the time, but
that cannot be counted on always to be present. All biological devices
are fallible devices, even when norrrlal and health^.'^
It follows that. the proximal assertability rules for a sentence
would not define its semantics, for they would not determine what
its truth conditions were. Rather, following proximal assertability
rules would be rneans that were, merely, approximations to the
end that was following correspondence truth rules-more or less
helpful and more or less reliable means to that end. Let us reflect.
for a moment. upon certain consequences of this model.
If proximal assertability rules were rules that.we followed only as
a more or less reliable means to following distal truth rules, then it
would not at least be obvious that those who shared a language in
the sense of having cornpetences to abide by the same truth rules
would have any need to share proximal assertability rules as well.
The rnale hoverfly follows the distal rule "If you see a female,
catch it" by following the proxirnal hoverfly rule. The male
housefly follows the same distal rule by tracking the female rather
than by plotting an interception path, employing different proximal
means to the same distal end. Now consider how many different
ways there are to make a map of' a city: for example, by
walking about with a yardstick, paper and pencil, by working from
2 9 F ~ ramplification of this very crucial tileme, see my "Thoughts
Without Laws" and "Biosemantics."
RUTH GARRETT MZLLZKAN
aerial photographs, by using surveyors' instruments, etc. Might
there not also be various ways to make sentences that map onto the
world in accordance with the same truth rules? Is there really any
reason to suppose that only one set of proximal assertability rules
could effect a reasonably reliable competence to conform to a
given set of distal correspondence rules? Consider, for example,
how many ways there are to tell whether a solution is acid or
whether it has iodine in it. Consider how many alternative visual
and tactile clues we use, on one occasion or another, to perceive
depth. And consider: were the proximal assertability rules that
Helen Keller used when she spoke English the same as those that
you use? If not, does it follow that she did not really speak English
after all?
Indeed, there is a sense, there is a way of individuating rules, in
which it is impossible for people to share proximal assertability
rules. Proximal assertability rules that I conform to correlate happenings
at the periphery of my nervous system or body with sentences.
Proximal assertability rules that you conform to correlate
happenings at the periphery of your nervous system or body with
sentences. For us to "share a set of proximal assertability rules"
could not, of course, be for me to purpose to correlate happenings
at the periphery of your body with my sentences. If I purposed to
do that, I would be purposing to conform to a distal rule, not a
proximal rule. We could "share proximal assertability rules" only
in the sense that our rules ran parallel. But it is not immediately
obvious what the point of running parallel to one another with
language might be. Why would you take any interest in the sentences
I uttered, if these correlated only with what was happening
at the ends of my afferent nerves? Only if the proximal assertability
rules that you and I used effected relatively reliable conformity
to the same distalcorrespondence rules would there be any
point in talking to one another. But if agreement is effected on the
distal level, what need would there be for agreement on the proximal
level? Hence what reason is there to assume, say with Quine,
that comparison of only proximal rules ought to yield determinate
translation between idiolects?30
30Forfurther discussion of the relation of proximal assertability rules to
truth rules see my "The Price of CorrespondenceTruth,"Nozis 20 (1986),
TRUTH RULES AND HOVERFLIES
VIII. CAUSESOF VERIFICATIONISTMYOPIA
Given a biological approach, then, there are reasons to think
that truth rules may be distal correspondence rules, hence that
classical truth conditions may do work for semantics. But Putnam
and Dummett claim that any such view is unintelligible. Why?
Although there are passages in both Dummett and Putnam that
could be given a less sympathetic reading, the reason is not (or at
least is not simply) that these philosophers take understanding to
be something that must transpire before consciousness. A more
explicit theme is that understanding a language is a practical
ability, constituted by a set of dispositions, in this case, learned responses:
"Now when someone learns a language, what he learns is
a practice; he learns to respond, verbally and nonverbally, to utterances
and to make utterances of his own" ( D ~ m m e t t ) ; ~ ~"language
understanding [is] . . . an activity involving 'language entry rules'
(procedures for subjecting some sentences to stimulus control),
procedures for deductive and inductive inference and 'language
exit rules' . . . " ( P ~ t n a m ) . ~ ~It follows, Putnam and Dummett now
agree, that if a language is characterized by certain truth rules, this
fact must be one that shows up in the speech dispositions of the
language users. And it follows that if there are no dispositions to
recognize correspondence truth conditions, sentences can not have
correspondence truth conditions.
Yutnam's phrase "language entry rules" is a reference to Sellars,
but, of course, many other central figures have also held that understanding
a language must yield to a dispositional analysis,
among them Quine, Davidson, many would say Wittgenstein, and,
in the philosophy of mind (re: inner language), the functionalists.
Despite this distinguished advocac:~,surely Kripke's remark about
pp. 453-468, and also LTOUC, especially Chapters 8 and 9. Proximal assertability
rules are close relatives of what I there called "intensions."(In
this essay I am not emphasizing that perception cliaracteristically is an
activity involving overt exploration, a fact that was in the foreground
when I spoke of intensions in Ľf'OBC. Thus the notion "proximal assertability
rules" is a somewhat duller tool than I intended "intensions"to be
in L?%)BC.)
31"Whatis a 'Theoryof Meaning? (11),"p. 82.
3'''Realis~nand Reason," p. 110.
R U7.H GARRETT MILLIKAN
illegitimate "equation of performance with correctness" is applicable
here. To be competent in a language involves that one
have a practical grasp of its truth rules. About that everyone
agrees. But "true" is clearly a normative notion. "True" is how my
sincerely uttered sentences are when they come out mght, when they
are, using Kripke's expression, as they "ought to be," when I
achieve what I purpose in sincerely uttering them. And no mere
set of dispositions, no mere performance, determines a measuring
"ought," a standard or norm. No set of dispositions, then, could
determine truth rules.
Nor is the normative ingredient in truth provided by the fact
that the dispositions that constitute competence in a language must
agree with a public norm. Compare games. Consider first a case in
which I intend to play the same game as the others do, say, the one
they call "chess," but I mistakenly play by different rules than the
others. This is a case of playing wrongly in the sense that I have
not played the game I intended, or, perhaps, the one others expected
me to. Similarly, if I intend to use the same language as the
others, but in fact adopt different truth rules, then I speak
wrongly, for I have not spoken the language I intended, or that
others expected me to. This is called "not knowing the language"
or "making mistakes in the language." Second, consider a case in
which I have no intention to play with the chess pieces as the
others do nor do the others expect me to. Then playing by different
rules is just playing a different game. It is neither playing
chess wrongly nor doing anything else wrongly. The linguistic parallel
to this is called "speaking a different language." But speaking
wrongly in the sense of speaking falsely is still a third possibility.
Speaking falsely is not just a way of being out of step, nor is itjust
marching to the beat of a different drummer. Suppose we call it a
"rule" of chess that you are supposed to checkmate your opponent.
Then speaking falsely is like failing to checkmate the opponent.
Better, it is like failing to pick up one straw without moving
the others when playing jackstraws. Just as learning the rules
better is not the cure for losing at chess orjackstraws, learning the
community's language better is not the cure for bad judgment.
And just as whether one succeeds at jackstraws, that is, at not
moving the other sticks, does not depend on any agreement with
the community, neither does whether one succeeds in speaking
TRUTH RULES AND HOVERFLIES
truth in one's language. To purpose to follow certain truth rules is
to set a standard for oneself-a standard that one may fail to meet.
It is because purposes set standards that "true" is a normative
notion and that no set of dispositions could determine truth rules.
Similarly, although Dummett and Putnam are right that semantic
meaning must be resident somehow in language competence, no
set of dispositions equals a competence. First, a disposition does
not express a competence unless it is a disposition informed by a
purpose. My disposition to fall if left unsupported is no competence,
nor is the hoverfly's disposition to chase birds. Conversely,
having a competence does not, in general, imply that one has any
particular dispositions. If I know how to A-say, to sharpen a drill
bit-it doesn't follow that I have a disposition to succeed in A-ing
if I try. Perhaps my hands are too cold, or the only grindstone
available is not the kind I am practiced at using, or you insist on
joggling my elbow. Though I know how to walk, sometimes I trip
when I try. Recall the hoverfly, who exhibits a competence whenever
he conforms to the distal hoverfly rule in a normal way, yet,
due to the inconstancy of conditions outside him, often does not
manage to conform to it at all. Nor are there specified conditions
under which a person must succeed in order to know how. If I can
only sharpen the bit using one sharpening tool whereas you know
how to use another, then normal conditions for exercise of my
ability to sharpen a drill bit will be different from normal conditions
for exercise of yours; each may fail where the other succeeds.
Knowing how to do A entails, at best, only that there are some
normal conditions under which one succeeds in doing A.
Now there is an evident reason why knowing how to A does not,
in general, entail having any simple disposition to succeed in
A-ing. The reason is that most know-how involves distal action, and
there is no such thing as a simple disposition to involvement with
anything distal. How one interacts with things at a distance always
depends upon what lies in between, on surrounding conditions.
Simple dispositions can concern only reactions to and actions upon
that which touches one or, perhaps, what is inside one. It follows
that to assimilate language competence to a set of dispositions directly
begs the question against distal truth rules. There is no need
for tortuous arguments to demonstrate that truth rules must then
be verificationist. On a dispositional account, to "grasp" cor-
R U?'H GARRETT MILLIKAN
respondence truth rules for each sentence in one's language would
be to have a "capacity . . . to evince recognition of the truth of the
sentence when and only when the relevant condition is fulfilled"
( D ~ m m e t t ) . ~ ~But if a "recognitional capacity" is a disposition, it
must be a disposition to respond to a proximal stimulus, there
being no such thing as a disposition to respond to something distal.
And dispositions to respond to proximal stimuli with sentences
could correspond, at best, to assertability conditions, certainly not
to distal correspondence truth conditions. Q.E.D.
Compare the hoverfly. Assuming that his insides are working
right, what he has a disposition to do is, at best, to conform to the
proximal hoverfly rule. Does it follow that he has no ability to
catch females?
It is significant, I think, how close the dispositional view of language
understanding is to the more classical view that understanding
takes place wholly within consciousness. On the classical
view, understanding must ultimately involve relations only to
things that touch the mind. On the dispositional view, understanding
still involves only what touches the mind or, say, the
nervous system. It is easy, then, to slip back and forth between two
ways of interpreting the Dummett-Yutnam attack upon realist
Yet what Kripke has shown is that neither view of language
understanding is a tenable view. Hence, whatever may be said for
or against the positive theory of rule following that I have offered,
the verificationist vision is surely unnecessarily nearsighted. If
Kripke (and Kripke's Wittgenstein) are right, then whatever the
status of rule following, we have no reason to think that the following
of correspondence truth rules is any more problematic than
is the following of verificationist truth rules.
On the other hand, perhaps what is most puzzling about the
following of any kind of language rules is how one could "know"
these rules without having a prior language, a prior way of
"meaning" or thinking of these rules. Yet surely even the medium
of thought, even whatever is currently before the mind or in the
33"Whatis a Theory of Meaning? (II),"pp. 80-81.
341will not attempt to prove that Dummett and Putnam themselves do
some sliding, but on Putnam, see LTOBC, Epilogue.
TRUTH RULES AND HOVERFLIES
head, stands in need of interpretation. Knowing the rules is not a
disposition, nor can it be explained in the end by reference to
prior representations of the rules. The biological account agrees
with both of these consideration^.^^
University of Connecticut
35Earlierversions of this paper were read at the University of Wisconsin
(Madison), Western Michigan University, the University of Maryland,
Trinity University, the University of New England (Australia), Australian
National University, Monash University and Vanderbilt University. I am
grateful to the members of these departments, to Margaret Gilbert and
Tohn Trover. and to unknown referees for The Philosobhical Review. forJ ,
helpful comments and suggestions.
You have printed the following article:
Truth Rules, Hoverflies, and the Kripke-Wittgenstein Paradox
Ruth Garrett Millikan
The Philosophical Review, Vol. 99, No. 3. (Jul., 1990), pp. 323-353.
Stable URL:
http://links.jstor.org/sici?sici=0031-8108%28199007%2999%3A3%3C323%3ATRHATK%3E2.0.CO%3B2-5
This article references the following linked citations. If you are trying to access articles from an
off-campus location, you may be required to first logon via your library web site to access JSTOR. Please
visit your library's website or contact a librarian to learn about options for remote access to JSTOR.
[Footnotes]
3
Metaphysical Anti-Realism?
Ruth Garrett Millikan
Mind, New Series, Vol. 95, No. 380. (Oct., 1986), pp. 417-431.
Stable URL:
http://links.jstor.org/sici?sici=0026-4423%28198610%292%3A95%3A380%3C417%3AMA%3E2.0.CO%3B2-L
15
In Defense of Proper Functions
Ruth Garrett Millikan
Philosophy of Science, Vol. 56, No. 2. (Jun., 1989), pp. 288-302.
Stable URL:
http://links.jstor.org/sici?sici=0031-8248%28198906%2956%3A2%3C288%3AIDOPF%3E2.0.CO%3B2-G
17
Biosemantics
Ruth Garrett Millikan
The Journal of Philosophy, Vol. 86, No. 6. (Jun., 1989), pp. 281-297.
Stable URL:
http://links.jstor.org/sici?sici=0022-362X%28198906%2986%3A6%3C281%3AB%3E2.0.CO%3B2-9
23
Thoughts Without Laws; Cognitive Science with Content
Ruth Garrett Millikan
The Philosophical Review, Vol. 95, No. 1. (Jan., 1986), pp. 47-80.
Stable URL:
http://links.jstor.org/sici?sici=0031-8108%28198601%2995%3A1%3C47%3ATWLCSW%3E2.0.CO%3B2-V
http://www.jstor.org
LINKED CITATIONS
- Page 1 of 2 NOTE:
The reference numbering from the original has been maintained in this citation list.
26
Knowing One's Own Mind
Donald Davidson
Proceedings and Addresses of the American Philosophical Association, Vol. 60, No. 3. (Jan., 1987),
pp. 441-458.
Stable URL:
http://links.jstor.org/sici?sici=0065-972X%28198701%2960%3A3%3C441%3AKOOM%3E2.0.CO%3B2-P
26
Individualism and Self-Knowledge
Tyler Burge
The Journal of Philosophy, Vol. 85, No. 11, Eighty-Fifth Annual Meeting American Philosophical
Association, Eastern Division. (Nov., 1988), pp. 649-663.
Stable URL:
http://links.jstor.org/sici?sici=0022-362X%28198811%2985%3A11%3C649%3AIAS%3E2.0.CO%3B2-2
30
The Price of Correspondence Truth
Ruth Garrett Millikan
Nos, Vol. 20, No. 4. (Dec., 1986), pp. 453-468.
Stable URL:
http://links.jstor.org/sici?sici=0029-4624%28198612%2920%3A4%3C453%3ATPOCT%3E2.0.CO%3B2-B
http://www.jstor.org
LINKED CITATIONS
- Page 2 of 2 NOTE:
The reference numbering from the original has been maintained in this citation list.